Introducing Genius Cortex

Genius’s VP of Development, Dominic, uses this session to walk through exactly what AI in Genius ERP means today versus where it’s headed — without the usual hype. He opens with the practical roadmap non-AI work that has to happen alongside it: security and compliance (CMMC, ITAR, Canada’s Controlled Goods Program), more automation building blocks (workflows, custom fields, REST APIs), and a long-overdue communication layer that tracks emails against ERP documents. From there he gets specific about what’s actually shipped: a manufacturing-scoped chatbot (launched in version 17.1) trained only on Genius’s own help content, Smart Filters that turn a plain-language request into a database query, and Smart Paste, which can already turn a copied email or spreadsheet into quote or order lines. He’s just as direct about what isn’t built yet — no drawing-based part matching out of the box, no general-purpose custom agents (though a dedicated team can build them on request) — and spends real time on data security, confirming customer data is stored in its own private vector database and isn’t used to train the underlying model. The back half of the session is a deep, honest look at AI-driven recommended reordering: how it’s meant to encode a purchaser’s tacit experience, what it takes to train it well, and where it can (and can’t yet) be trusted. If you want a realistic sense of what Genius’s AI can do right now versus what’s still on the roadmap, this is a good primer.

What You’ll Learn In This Webinar

Chapter 1 — Introduction to the Webinar

Frank (11 years at Genius, hosting monthly webinars for 5) introduces Dominic, who’s led Genius’s development for 25 years, for a session on AI for manufacturers. Genius is framed with a few quick facts: North America-wide, 140 employees, 30+ years serving manufacturers, and a 96% customer retention rate.

Chapter 2 — Unique Features of Genius ERP

Genius is built specifically for engineer-to-order, configure-to-order, and make-to-order manufacturers — high-mix, low-volume environments. Frank points to the CAD2BOM integration, an implementation team with deep niche experience, and built-in business intelligence as differentiators, noting the platform covers everything from scheduling and estimating to field service, warranties, and claims tracking as core functionality.

Chapter 3 — Team Introduction and AI Transition

Dominic introduces Genius’s 60+ person product team — product owners (Martin, Alex, Robert) who manage the backlog, an architecture lead (Mohammed), a team manager (Jeremy), and a release orchestrator — before noting the newest additions: data scientists and natural language programmers, reflecting Genius’s own internal shift toward AI skills.

Chapter 4 — Roadmap for Security and Compliance

Beyond AI, Dominic outlines three near-term priorities: security/compliance (working toward CMMC, ITAR, and Canada’s Controlled Goods Program requirements within two years), automation building blocks (workflow steps, custom fields, per-action security, REST APIs), and a new communication layer — an Outlook plugin that tags incoming emails to Genius transactions and tracks that history per document.

Chapter 5 — Long-Term Vision for AI in Manufacturing

Alongside the two-year roadmap, Dominic describes a 10-year vision exercise: a roundtable of customers, partners, and manufacturing/technology experts brought in to picture the future and stress-test Genius’s roadmap. Pilot customers (including Meshtech) helped validate real-world value before wider release, alongside two partners — a technology partner supporting AI development and training, and Scale AI, a Canadian consortium helping fund and scale the effort.

Chapter 6 — Introduction of AI Features in Genius

The chatbot, launched with version 17.1, is trained on Genius’s help files and Academy content and deliberately scoped to manufacturing context only — not the open internet — and works in both English and French. Looking ahead, next year’s version will let customers add their own knowledge (coding conventions, quality manuals), with a longer-term vision of the chatbot becoming a central place to ask questions and eventually execute transactions directly.

“We’re not just taking OpenAI, plugging it there, making it available to you guys, and then having fun with it.” — Dominic

Chapter 7 — Chatbot Capabilities and Future Enhancements

Dominic addresses data security directly: personal data is protected, and customer information is stored in its own private vector database rather than being shared with other customers or used to train the underlying AI model — a commitment that will be part of Genius’s SLA.

Chapter 8 — Data Security and AI Training Concerns

A wide-ranging live Q&A: drawing-based part matching isn’t available out of the box today (though technically possible in the future); AI hallucination does happen and is managed through dedicated quality and test plans; semantic search understands intent and translation, not just exact wording; Smart Filters turn plain-language requests into queries (and can be saved for reuse); a dedicated Genius team can build custom AI agents on request even though it’s not a packaged feature yet; Smart Paste is already available for quotes and orders, turning copied email or spreadsheet text into transaction lines; and the planned Outlook plugin would detect a contact, match it to an existing quote or create a new one, match attachments to line items, and centralize communication history on any document type — available for both SaaS and on-premise customers.

Chapter 9 — AI for Predictive Reordering

Recommended reordering is an existing module getting new AI-driven screens rather than a new concept — but Dominic is candid that it only works well when underlying data (inventory levels, min/max, lead times, preferred vendors) is accurate, which often isn’t the case in practice, leading experienced purchasers to over-order or compensate manually. The goal is to encode that tacit “trained purchaser” knowledge into the AI. Using real historical data from five customers, Genius backtested several AI models against actual purchaser decisions on lead-time predictions and selected the best-performing model — finding purchasers tend to react reactively and inconsistently to problems, while the AI takes a steadier, higher-level view, provided the underlying customer data and behavior is good to begin with.

Chapter 10 — User Interaction with AI Predictions

Each item gets its own demand-pattern model — punctual, seasonal, trending, or intermittent — and the resulting chart shows historical demand alongside the AI’s forecast and a confidence band, so a well-established item might get a high-confidence recommendation while a one-off purchase gets a clearly flagged low-confidence one. Buyers can chat directly with the AI about a specific recommendation to understand its reasoning, and if a recommendation proves reliable over time, it can eventually be automated by default — but the buyer always retains control over whether to activate it.

Chapter 11 — Upcoming Webinars and Future Topics

Frank previews a Nov. 20 customer-specific session on what’s new in Version 17, followed by a Dec. 2 podcast-style conversation with Genius’s head of professional services on post-implementation continuous optimization, and an open-mic follow-up the week after. In closing Q&A: recommended reordering based on min/max already exists today, with AI aimed at flagging when the underlying data itself is inaccurate; drawing-to-cycle-time estimation is handled today through existing software partnerships rather than natively; and automation questions (REST API, Power Automate, existing automated tasks) point back to a past webinar on automation, with an open invitation to email customer success with specific project scopes.